A Non-Cooperative Satellite Feature Point Selection Method for Vision-Based Navigation System.

Area of Science:

• Spacecraft Guidance, Navigation, and Control (GNC)
• Computer Vision and Image Processing
• Robotics and Autonomous Systems

Background:

• Monocular vision-based relative navigation for non-cooperative targets is crucial for space missions.
• The number of feature points significantly impacts onboard computational load.
• Efficient feature point selection is needed to balance accuracy and computational cost.

Purpose of the Study:

• To propose a quasi-optimal method for selecting a geometrically well-distributed subset of feature points.
• To reduce the computational load in monocular vision-based relative navigation.
• To verify the performance of the proposed method using Dilution of Precision (DOP).

Main Methods:

• A quasi-optimal feature point selection method based on geometric distribution and redundancy calculation.
• Defining cost as a function of the angle between unit vectors to feature points.
• Iteratively removing the most redundant feature point until the desired subset size is reached.
• Utilizing the Dilution of Precision (DOP) concept to evaluate performance.

Main Results:

• The quasi-optimal method successfully selects feature points with good geometric distribution.
• Simulation results demonstrate the feasibility of calculating relative position and attitude with the selected subset.
• The selected feature points provide high accuracy with significantly reduced computation time.

Conclusions:

• The quasi-optimal method is effective for feature point selection in vision-based satellite navigation.
• This method offers a practical solution for reducing computational load without sacrificing accuracy.
• The approach enhances the efficiency of onboard processing for relative navigation tasks.